JP4694126B2 - Process for producing hydrocarbon products having a sulfur content of less than 0.05% by weight - Google Patents

Description

FIELD OF THE INVENTION The present invention starts with two or more hydrocarbon feedstocks (optionally including other feedstocks) having a sulfur content greater than 0.05% by weight and has a desired sulfur content of 0.05. It relates to a process for continuously producing a hydrocarbon product specified to be less than% by weight. The invention particularly relates to a process wherein the hydrocarbon product is a gas oil (diesel) product.

BACKGROUND OF THE INVENTION A so-called refinery process in a so-called gas oil blending repository where various hydrotreated or non-hydrotreated gas oil components stored in a storage vessel are blended to obtain a finished gas oil product. Is known. These hydrotreated gas oil components are obtained by treating various sources of suitable hydrocarbon refinery streams with high sulfur content in a hydrodesulfurization process unit to reduce the sulfur content to low levels. It is done. Examples of such refinery streams are kerosene fractions, straight run gas oil, vacuum gas oil, gas oil obtained by thermal cracking, and light and heavy circulating oil obtained by fluid catalytic cracking units. An example of a non-hydrotreated gas oil component used in the production of finished gas oil with this blending method is a gas oil fraction obtained in a fuel hydrocracking process.

In recent years, the standard value of sulfur for gas oil has fallen sharply from environmental requirements. Such further decrease in the specific sulfur level is expected. When it is necessary to meet such tight sulfur specifications, the above method of producing gas oil within specifications is not fully satisfactory for maximizing gas oil production.
WO-A-0029512 EP-A-58383 WO-A-011116 WO-A-011117 WO-A-0183406 WO-A-0183648 WO-A-0183647 WO-A-01833641 WO-A-0020535 WO-A-0020534 EP-A-1101813 US-A-5888376 US-A-6204426 GB-A-2309918 EP-A-555216 Handbook of Petroleum Refining Process, Editor-in-Chief Robert A. Meyers, 2nd edition, McGraw Hill, 8.29, pages 8.38. Field F. Hupkes W. Inferred Measurement, Hydrocarbon Engineering, April 2001, pp. 73-76. Introduction to linear regression analysis by Montgomery and Peck, John Willy & Sons, 1992 Linear Systems by Keilath, Prentice-Hall, Information & Systems sciences series, 1980 T.A. Poggio and F.M. Girosi, Network for application and learning, The Proceedings of the IEEE, 78 (9): 1485-1497, September 1990. “Stochastic Progressing and Filtering Theory” by Ajazwinski (Academic Press, Mathematical Sciences and Science and Engineering, Vol. 64, 1970)

  It is an object of the present invention to provide a method that can more easily maximize the production of low sulfur gas oil in a refinery environment.

Detailed description of the invention This object is achieved by the following method. That is, the process starts with two or more hydrocarbon feedstocks having a sulfur content greater than 0.05% by weight, and a hydrocarbon product having a desired sulfur content specified less than 0.05% by weight. In a method of continuously producing
(A) a step of blending two or more hydrocarbon feeds having a sulfur content exceeding 0.05% by weight to form a blended raw material mixture;
(B) a step of reducing the sulfur content of the blended raw material mixture in the hydrodesulfurization step;
(C) After obtaining the hydrocarbon fraction with reduced sulfur content, including the effluent of step (b), measuring the sulfur content of the hydrocarbon fraction, and (d) direct production of step (c) After obtaining the final hydrocarbon product from the product, the sulfur content measured in step (c) is compared with the desired sulfur content of the hydrocarbon product, and the sulfur content of the hydrocarbon product is Adjusting the process until it approximates or is equivalent to the desired sulfur content of the hydrogen product;
Optimized the production of hydrocarbons with the desired sulfur content by integrated control of the blending in step (a) and hydrodesulfurization unit operation in step (b) and measured in step (c) The sulfur content of the hydrocarbon fraction is taken into account.

  It has been found that according to the method of the present invention, a method for producing a low sulfur gas oil with improved control of the low sulfur gas oil product can be obtained. Such process control and improved line-up has made it possible, for example, to maximize low sulfur gas oil production. A further advantage is that no compound storage is required. The hydrocarbon fraction obtained as a direct product of step (c) forms a hydrocarbon product directly or after accumulation in the storage vessel. In the following, the present invention will be described in detail with some preferred embodiments. Further advantages of the present invention will become apparent from this description.

  The process of the present invention is suitable for use in any process for producing low sulfur products, such as automobile gasoline, but is particularly suitable for the production of gas oil products. Gas oil products are also referred to as diesel products in the United States, but the volume% recovered at 250 ° C is preferably less than 65% (V / V) and the 95% point is preferably less than 360 ° C The cetane index is preferably greater than 40 or greater than the corresponding cetane number, the cloud point is preferably less than 0 ° C., and the polyaromatic hydrocarbon content is preferably 11% (m / m) and the flash point is preferably more than 55 ° C. Although the following description will often refer specifically to gas oils, it should be understood that the techniques disclosed below are applicable to the production of other low sulfur refinery products in accordance with the spirit of the present invention.

The feedstock of the process is two or more hydrocarbon feedstocks with a sulfur content exceeding 0.05% by weight. In addition, other feeds such as low sulfur content hydrocarbon feeds and additives may be used.
In step (a), two or more hydrocarbon feedstocks having a sulfur content exceeding 0.05% by weight are blended. Preferably 3 or more, more preferably 4 or more of these high sulfur feedstocks are blended in step (a). An advantage of the present invention is that the more the high sulfur feedstock is blended in step (a), the easier it is to achieve the present invention. With such a large number of feedstocks, the method of optimizing the entire process becomes increasingly complex. It has been found that optimization improvements can be achieved by the following preferred method.

  Examples of feedstocks used in step (a) are typically various refinery sources, namely kerosene fraction, straight run gas oil, vacuum gas oil, gas oil obtained by pyrolysis, and fluid catalytic cracking unit Is a light or heavy circulating oil obtained in The kerosene fraction according to the invention has an initial boiling point of 160-180 ° C and a final boiling point of 220-260 ° C. The straight run gas oil fraction is a gas oil fraction obtained by atmospheric distillation of crude oil refinery feedstock. The straight boiling gas oil fraction has an initial boiling point of 180-280 ° C and a final boiling point of 320-380 ° C. The vacuum gas oil is a gas oil fraction obtained by vacuum distillation of a residue obtained by atmospheric distillation of the crude oil refinery feedstock. The initial boiling point of vacuum gas oil is 240-300 ° C, and the final boiling point is 340-380 ° C. The pyrolysis method also produces a gas oil fraction that can be used in step (a). The gas oil fraction has an initial boiling point of 180-280 ° C and a final boiling point of 320-380 ° C. The initial boiling point of the light circulating oil obtained by the fluid catalytic cracking method is 180 to 260 ° C, and the final boiling point is 320 to 380 ° C. The initial boiling point of the heavy circulating oil obtained by the fluid catalytic cracking method is 240 to 280 ° C, and the final boiling point is 340 to 380 ° C. The sulfur content of these feedstocks exceeds 0.05% by weight. The maximum sulfur content is about 2% by weight.

  In step (a) of the method of the present invention, various high sulfur feedstocks obtained from various refinery sources are blended. Any excess feedstock that cannot be used directly in step (a) can be used as a feedstock for refinery hydroconversion units, such as fuel hydrocracking units or fluid catalytic cracking units, or temporarily stored in storage containers it can. The contents of this storage space, which can contain one or more storage containers, can later be used as a compounding ingredient. Two or more storage vessels are preferably used, one used for storage of any surplus circulating oil and the other storage vessels can be used for any mixture of remaining hydrocarbon feedstock. The circulating oil is preferably placed away from other feedstocks.

Step (b) is preferably performed in the state of a conventional hydrodesulfurization unit. In such units, the blended feed mixture is contacted with a suitable hydrodesulfurization catalyst in the presence of hydrogen in the reactor. Sulfur component reacts H ₂ S becomes together with other light components of the finishing (work-up) and the discriminating section hydrodesulfurization unit is easily removed from the hydrodesulfurization unit reactor effluent. The catalyst is preferably a heterogeneous catalyst containing a support, a Group VIB metal and a Group VIII non-noble metal. Examples of suitable catalysts are nickel-molybdenum catalysts supported on alumina or cobalt-molybdenum catalysts supported on alumina. A possible hydrodesulfurization process is described in Handbook of Petroleum Refining Processes, Chief Editor Robert A. Meyers, 2nd edition, McGraw Hill, 8.29, 8.38.

  In step (b), further catalytic dewaxing of the gas oil product can be advantageously carried out in order to reduce the pour point and / or cloud point of the gas oil product. The catalytic dewaxing is preferably performed after the hydrodesulfurization step. More preferably, the hydrodesulfurization step and the catalytic dewaxing step are carried out in one reactor in which different catalysts are present in a series of stacked beds. Suitable dewaxing catalysts contain molecular sieves, binders and Group VIII metals, preferably non-noble metals such as nickel or cobalt. The molecular sieve material is typically an intermediate pore size molecular sieve having a pore size in the range of 0.35 to 0.80 nm. Suitable molecular sieves are ZSM-11, ZSM-12, ZSM-22, ZSM-23, ZSM-35, ZSM-48, ZSM-57, SSZ-23, SSZ-24, SSZ-25, SSZ-26, SSZ-32, SSZ-33 and MCM-22. Preferred molecular sieves are ZSM-5, ZSM-12 and ZSM-23. The binder is preferably a low acidity refractory oxide binder material that is essentially free of alumina, preferably silica. The surface of the aluminosilicate zeolite crystallites exemplified above is preferably modified by surface dealumination treatment. Furthermore, such dewaxing catalysts, their production and their use in gas oil dewaxing are described in WO-A-0029512.

  In step (c), the effluent of step (b) is preferably mixed with a low sulfur hydrocarbon feedstock having a sulfur content of less than 0.05% by weight, if available from a particular refinery. For example, a refinery equipped with a fuel hydrocracker will produce a gas oil feedstock with a low sulfur content of less than 0.05% by weight in the hydrocracker. Other examples of such feedstocks are fuel fractions obtained from hydrotreated kerosene fraction, fatty acid methyl ether, and Fischer-Tropsch reaction products. It is clear that it is advantageous to blend these low sulfur fractions in the effluent of step (b) rather than blending them in the high sulfur fraction in step (a).

  When a Fischer-Tropsch derived gas oil is added, this gas oil is preferably obtained from a (hydrocracked) Fischer-Tropsch synthesis product. Examples of Fischer-Tropsch derived gas oils are EP-A-583836, WO-A-011116, WO-A-011117, WO-A-0183406, WO-A-0183648, WO-A-0183647, WO-A-. No. 0833641, WO-A-0020535, WO-A-0020534, EP-A-1101813, US-A-5888376 and US-A-6204426.

  The Fischer-Tropsch derived gas oil suitably consists of at least 90% by weight, more preferably at least 95% by weight iso and linear paraffin. The weight ratio of isoparaffin to normal paraffin is preferably above 0.3. This ratio may be 12 or less. Preferably, this ratio is 2-6. The actual value of this ratio is measured in part by the hydroconversion method used in the production of Fischer-Tropsch derived kerosene or gas oil derived from the Fischer-Tropsch synthesis product. Some cyclic paraffin may be present.

Suitably, the cetane number of the Fischer-Tropsch derived gas oil is greater than 60, preferably greater than 70, and the distillation curve is largely within the normal gas oil range of about 150-400 ° C. The Fischer-Tropsch gas oil preferably has a T90 wt% of 340-400 ° C, a density of about 0.76-0.79 g / cm ³ at 15 ° C, and a viscosity of about 2 at 40 ° C. .5 to 4.0 cSt.

  The additive is preferably added to the effluent of step (b) in step (c). Examples of gas oil additives increase cetane number, adjust electrical conductivity, suppress room temperature flow characteristics such as cold filter clogging point and pour point, and / or improve color tone and lubricity Is an additive.

  The sulfur content of the hydrocarbon fraction obtained in step (c) is measured. This measurement can be performed by an on-line analyzer or an off-line measurement which is measured by, for example, X-ray fluorescence or ultraviolet fluorescence. Alternatively, the sulfur content can be measured using a near infrared measurement method, for example as described in GB-A-2309918. In addition, a model-based quality estimator as described in more detail below can be used to measure the sulfur content.

  In step (d), the sulfur content measured in step (c) is compared with the desired sulfur content. If the sulfur content measured in step (c) is too different from the desired sulfur content of the final gas oil product, this method needs to be adjusted. Adjustment of this method is performed by the integrated control of the blending in step (a) and the hydrodesulfurization unit operation in step (b). This integrated control will be described in more detail below, but preferably by adjusting the operating conditions of the hydrodesulfurization unit in step (b) and changing the composition of the blended raw material mixture obtained in step (a). It is to adjust the characteristics of the blended raw material mixture. It can also be adjusted by optionally adding a low sulfur hydrocarbon feed in step (c).

  The adjustment of the process operating conditions for the hydrodesulfurization unit in step (b) is preferably performed using a model-based controller, such as a multivariable controller, in particular a well-known multivariable predictive controller. Artificially manipulated process conditions to achieve the desired sulfur content are, for example, feed rate of the blended feed mixture to the hydrodesulfurization unit, hydrogen recycle and the temperature distribution of the hydrodesulfurization reactor. Hydrodesulfurization reactor temperature distribution can be affected by adjusting the feed inlet temperature or by adjusting the amount of quench mixture added to the reactants between the two catalyst beds of the hydrodesulfurization reactor. be able to. It may be advantageous for the quench mixture to be part of the blended raw material mixture obtained in step (a). Such a control loop of course also takes into account the associated limitations in the hydrodesulfurization unit. If the sulfur content of the final gas oil product approximates or is equal to the desired sulfur content, the multivariable predictive controller preferably maximizes the feed rate to the hydrodesulfurization unit.

  The sulfur content measured in step (c) is preferably the sulfur content of the gas oil fraction after adding some low sulfur feedstock. This is advantageous because it results in the final gas oil fraction of step (c), the sulfur content of the direct product of step (c), which determines the sulfur content of the final gas oil product, ie the hydrocarbon product. is there. For example, if a large amount of low sulfur feedstock for blending can be temporarily used to blend into the gas oil effluent of step (b), the sulfur content in the direct product of step (c) after blending is The reduction in the required amount of sulfur in the hydrodesulfurization step of step (b) can be mitigated (that is, the reduction in the required amount of sulfur is reduced) so that the state approximates the desired value. This allows more blended raw material mixture to be processed in step (b), or more of the high sulfur gas oil feedstock, such as circulating oil, is one of the blended feed mixture produced in step (a). Can be a part.

  The sulfur content of the final gas oil product approximates or is equal to the desired sulfur content. By approximation is meant the desired sulfur content, preferably within 10 ppm (0.001 wt%), more preferably within 5 ppm. The desired sulfur content is less than 0.05% by weight, preferably less than 0.01% by weight, more preferably less than 0.005% by weight of products that may vary in different market conditions. Depends on the standard value. This desired sulfur content usually exceeds 0.0005% by weight (5 ppm), but this lower limit may even be lowered if tighter administrative regulations are enforced.

  The final gas oil product may be the direct gas oil fraction obtained in step (c). Here, the sulfur content of the gas oil fraction obtained in step (c) means that it approximates or is equal to the desired sulfur content for a certain period of time. The point at which the desired sulfur content is required is when, for example, the fraction obtained in step (c) is transported directly to the product pipeline or loaded into a transport means such as a ship or train.

  Alternatively, the gas oil fraction of step (c) may be first collected and stored in a storage container. In this case, it is important that the sulfur content of the mixture in the storage container after filling to a predetermined level approximates or is equal to the desired sulfur content. The properties of this final gas oil product in the storage vessel will eventually calculate the raw material fed to the storage vessel, i.e. the average properties of the effluent of step (c) and the quality of the material present in the storage vessel at the beginning of operation. This can be derived. Therefore, in this case, the comparison between the measured sulfur content of step (c) in step (d) and the desired sulfur content of the final gas oil product is first made by measuring the sulfur content in step (c). The sulfur content of the gas oil already present in the storage container is estimated on the basis of the average sulfur content based on the above, and then the sulfur content estimation of the gas oil in the container is determined as a desired sulfur content. This is done by comparing. In such a case, the sulfur content of the hydrocarbon (gas oil) fraction obtained in step (c) remains constant approximate to the desired sulfur content while being supplied to the storage vessel. Not very important. If the process control can also compensate for the low sulfur content of the existing mixture in the storage vessel or later generated hydrocarbon fraction, the sulfur content of the hydrocarbon fraction can exceed the desired sulfur content for a period of time. . This means that the process is more adaptable to obtain the desired final stored gas oil product.

  In order to adjust the required sulfur content ("controlled variable") measured in step (c), a sulfur content estimate of the partially filled storage vessel is used. For example, if the sulfur content of the existing product in the storage container is less than the desired sulfur specification value, the required sulfur content in step (c) may be relaxed (ie, a higher sulfur content). Thus, using an estimate of the quality of the storage tank and using this estimate, the required sulfur content reduction in step (b), the composition of the blended feedstock mixture in step (a), and / or step (c) The method of mitigating or tightening the effects of the low sulfur feedstock formulation at) further reduces the likelihood of the excess quality (ie, higher than required quality) product.

  Between the operating conditions of the hydrodesulfurization unit in step (b) and the response of the sulfur analyzer to the hydrocarbon fraction obtained in step (c) there is usually a significant change in downtime. Due to this dead time, this response time before the hydrodesulfurization unit is adjusted with a multivariable predictive controller, for example by changing the raw material composition, becomes important. This produces, for example, non-standard products or non-optimized gas oil. For this reason, it is preferable to use a model reference quality estimator for measuring the sulfur content of the fraction in step (c). The performance of the multivariable predictive controller is that when the sulfur content predicted by the model reference quality estimator is used as a so-called “controlled variable” in the multivariable predictive controller instead of the measurement signal of the analyzer as described above, improves. Such a model reference quality estimator needs other information to predict the sulfur content. Such other information includes the sulfur content of the blended raw material mixture obtained in step (a), the sulfur content of the low sulfur feedstock that can be added in step (c), and the operating conditions of the hydrodesulfurization unit, for example It may be the partial pressure of hydrogen, the average reactor temperature, and / or the hydrodesulfurization operating conditions described above.

  In addition to measuring the sulfur content of the fraction obtained in step (c), the model reference estimator can also be advantageously used to predict some other relevant gas oil properties for this fraction. Examples of these properties are the cetane index, cetane number, cloud point, normal oil clogging point, flash point, pour point, density, viscosity, color tone, lubricity, electrical conductivity, total aromatic content, di + Aromatic content, poly-aromatic content, distillation temperature at 90%, 95% or 100% recovery, distillation curve, distribution by sulfur species boiling range, and nitrogen content. Such a model reference quality estimator is preferably used as an input for the properties of the blended raw material mixture, hydrodesulfurization operating conditions and / or properties and volumes of the fractions and additives added in step (c). . These estimated characteristics are preferably used in a more sophisticated control method as described below. If the final gas oil product is obtained in the final storage container, the other remaining relevant gas oil properties of this storage product can be estimated as well as in the case of sulfur. These estimates of stored gas oil product characteristics can be used in more sophisticated control methods as described below.

  Model reference quality estimators are well known, e.g. Hupkes W. Inferred Measurement, Hydrocarbon Engineering, April 2001, pages 73-76. Such a model reference quality estimator is preferably calibrated from time to time. The calibration is preferably performed using an effective real and measured measurement of the performance estimated by the quality estimator. Calibration is usually performed under steady state conditions in order to compare the effective actual measurement with the estimated value. Such a comparison is difficult to perform under non-steady state conditions when there is significant downtime as described above. Applicants have now found a way to eliminate these problems and calibrate the model reference quality estimator (QE) online under non-steady state conditions. This so-called robust quality estimator (RQE) is preferably used according to the method of the present invention. The novel method of robust quality estimator is described in detail below (see, eg, FIGS. 3 and 4). An effective actual measurement may be a laboratory analysis, more preferably a laboratory analysis with an on-line analyzer that can use near infrared (NIR) or nuclear magnetic resonance (NMR) spectroscopy.

  In step (a), in order to obtain a blended raw material mixture, the blending ratio of the plurality of high sulfur feedstocks is controlled. This blending ratio is compatible with the specific sulfur specifications of the blended raw material mixture in step (a), and this blended raw material mixture, the effluent in step (b), the hydrocarbon fraction in step (c) and / or Finally, it is preferred to select such that one or more other characteristics of the final hydrocarbon product fall within the desired specifications. Such other performances include: cetane index, cetane number, cloud point, normal oil clogging point, flash point, pour point, density, viscosity, color tone, lubricity, electrical conductivity, total aromatic content, di + It may be aromatic content, poly-aromatic content, distillation temperature at 90%, 95% or 100% recovery, distillation curve, distribution by sulfur species boiling range, and nitrogen content.

  In step (a), the sulfur content can be measured. More preferably, the blending of step (a) is controlled using a model-based quality estimate of the sulfur content in the blended raw material mixture. Such blending operations based on the sulfur content of the blended raw material mixture can still result in possible excess quality (ie, higher than required) in the remaining gas oil characteristics. In order to reduce the excess quality and increase the robustness of the operation, a known so-called compound performance controller is preferably used in step (a). Based on the characteristics of the various compounding ingredients (ie, the hydrocarbon feedstock used in step (a)) and the economic price of these compounding ingredients, the compounding performance controller determines how to make the compound to the desired characteristics. And can be optimized for the lowest cost. The formulation performance controller controls the formulation method based on the quality (sulfur content and / or one or more other characteristics) of the blended raw material mixture obtained in step (a). These characteristics can be directly measured or estimated by a model reference quality estimator, more preferably by the robust quality estimator. The input to the estimation model as used in the quality estimator is preferably the blending ratio, performance blending rules and / or blending inventory of the various feedstocks used in step (a). Various feedstock properties can be measured online or offline. A preferred method for measuring various properties is a method using near infrared, as described in GB-A-2303918 and EP-A-555216, for example.

  The model reference quality estimator must be calibrated from time to time to compensate for model inaccuracies or drift. Effective measurements of the properties estimated by the model reference estimator can be measured with offline laboratory sampling or offline semi-automatic near infrared / NMR analyzers. It is preferable to use advanced statistical process control techniques to check whether the quality estimator should be updated. In a further preferred embodiment, the robust quality estimator is used to estimate the properties of the blended raw material mixture.

  The control plan, optionally combining a (robust) quality estimator with the control loop of the multivariable prediction controller and the compound performance controller, is only when there is an integration between the control of the compound performance controller and the multivariable prediction controller. Suitable control is obtained. In this control plan, the sulfur content is controlled by a multivariable predictive controller, while other characteristics are controlled by a formulation performance controller. Without integrated control, a collision occurs between the multivariable predictive controller and the formulation performance controller. In order to resolve these conflicts, the integrated control preferably includes a global reconciliation layer (e.g., rules governing multivariable predictive controllers and formulation performance controllers). Without an overall adjustment layer, for example, multivariable predictive controllers may reduce hydrodesulfurization incorporation to ensure that sulfur standards meet when hydrodesulfurization faces dynamic constraints. . However, from an economic point of view, in these special situations, the hydrodesulfurization feed mix was adjusted to help the multivariable predictive controller to achieve sulfur control while maximally incorporating hydrodesulfurization. There is a possibility that this is more advantageous. The overall adjustment layer avoids solutions that are less than such optimal control.

More preferably, both control of the blending operation and the hydrodesulfurization unit are incorporated into one expanded multivariable predictive controller. Such a controller optimizes both the blending operation in step (a), the hydrodesulfurization process in step (b), and optionally the final product blend in step (c). Preferably, this maximizes economic benefits. An example of a suitable augmented multivariable predictive controller is Marquis P.M. Boustail J .; P. , Shell Multivariate and Optimiser Controller (SMOC), a bridge between State Space and Model Predictive Controllers, Application to the automation of a hydrotreating unit, IFAC Model Based Process Control, Georgia, USA, 1988 years, described in detail on pages 37-45 SMOC (Shell Multivariate and Optimized Controller). In such a control structure, the augmented multivariable predictive controller determines the sulfur content of the fraction obtained in step (c) and preferably the remaining one or more gas oil properties (ie, “controlled variables”). , Blending operation in step (a), optionally blending the effluent of step (b) in step (c) with a low sulfur hydrocarbon feed, and hydrodesulfurization in step (b) as described above. Control the unit immediately by skillfully manipulating the control of the unit. This expanded multivariable predictive controller also optimizes the entire process from an economic point of view by maximizing the overall nonlinear profit function (“maximum profit”). This non-linear profit function is expressed as component / product price per weight or volume unit (“$ component Includes an online adjustable factor for i "). The maximum profit is
Maximum profit = product Flow rate * $ Product- (Component i Flow velocity * $ component i) total

Where “product” “Flow rate” is the flow rate (weight or volume per hour) of the finished product in step (c), and “$ product” is “$ component” The price of the finished product using the same unit conversion as in “i” i The “flow rate” is the “product” of the hydrocarbon feedstock in step (a) and optionally in step (c). It is the flow rate in the same unit conversion as in the case of “flow velocity.” Maximizing the profit function is the same as minimizing the cost function. The latter can be expressed by total conversion of the above formula.

  The sulfur and other one or more gas oil properties of the hydrocarbon fraction obtained in step (c), and the properties of the blended raw material mixture obtained in step (a) are estimated multi-variable predictive controllers. To be supplied. The model reference quality estimator, more preferably the robust quality estimator, is used to estimate the quality (characteristic) of the gas oil.

The present invention will be described below with reference to FIGS.
FIG. 1 is an explanatory diagram of the state of the hydrodesulfurization technology team.
FIG. 2 shows an embodiment with a preferred control scheme of the present invention.
3 and 4 show a robust quality estimator.
FIG. 1 shows the state of the hydrodesulfurization operation technique. Various raw materials (102, 103, 104) are supplied to the hydrodesulfurization unit (101). These raw materials are stored in storage containers (105, 106, 107). Such storage is arranged between various hydrocarbon feedstocks (not shown) and the hydrodesulfurization unit (101). The product of the hydrodesulfurization unit is stored in different containers (108, 109, 110) of the compounding store (111). By storing the low sulfur hydrodesulfurization product away from the relatively high sulfur content hydrodesulfurization product, the gas oil product stored in the storage vessel (112) having the desired specifications can be formulated. . The compounding repository (111) may comprise a storage vessel (113) for low sulfur hydrocarbon feedstock that is not the product of a hydrodesulfurization unit. Additives are added to the final gas oil product by online additive injection (114).

  A preferred embodiment of the present invention of FIG. 2 will be described. FIG. 2 illustrates a process for continuously producing a gas oil product (120) having a desired sulfur content. In the illustrated methods, the kerosene feedstock (121), straight-run gas oil feedstock (122), vacuum gas oil feedstock (123) and circulating oil feedstock (sulfur content exceeding 0.05% by weight) are used. 124). According to step (a) of the present invention, a selected mixture of these feedstocks is formed to obtain a blended raw material mixture (125). Any excess feedstock can be temporarily stored in storage containers (126, 127). A portion of the feedstock already stored may be part of the blended raw material mixture. The blending operation is operated using a valve (128). These valves are controlled via the control line (130) by the enlarged model process controller (129). In order to control the blending operation, the quality of the blended raw material mixture (125) is estimated using a robust quality estimator (131). The estimated quality is at least a sulfur content and preferably one or more other gas oil properties.

  In step (b) of this method, the sulfur content of the blended raw material mixture is reduced in the hydrodesulfurization unit (132). The operating conditions of the hydrodesulfurization unit (132) are expanded so that the amount of sulfur is sufficiently reduced in this unit (132) to obtain a final gas oil product (120) having the desired sulfur and remaining properties. It is controlled via (136) by the model process controller (129).

  In step (c) of this method, the low sulfur gas oil feedstock (134) obtained in the hydrocracker is added to the effluent (133) of the hydrodesulfurization unit (the effluent of step (b)). This feedstock is fed from another feedstock storage container (135). The speed of the gas oil feedstock (134) is controlled via (137) by the enlarged model process controller (129). The effluent (133) is supplied with additive from one or more additive storage containers (139) by an in-line additive injector (138). The addition rate of the additive may be controlled by the controller (129) or separately. After the addition of low sulfur feedstock and additives, the sulfur content and preferably one or more other properties of the resulting hydrocarbon fraction, which is the direct product of step (c), are measured. The

  The direct product of step (c) thus obtained may be collected and stored in the final gas oil product storage container (140) or may be loaded directly on, for example, a ship (141). The sulfur content and one or more remaining characteristics of the hydrocarbon fraction obtained in step (c) and fed to the storage vessel (141) are estimated using a robust quality estimator (142). be able to. These estimates are the “controlled variables” used in the controller (129). The characteristics of the gas oil product in the storage tank (140) are also estimated using the robust quality estimator (143). These estimates are also used as inputs to the controller (129). Based on these values, the optimizer can adjust the set point for the sulfur content of the intermediate product (and the remaining “controlled variables”) fed to the storage vessel (140).

In situations where step (c) product quality needs to be continuously approximated to the desired quality of the final gas oil product, such as when loaded on a ship (141), the robust quality estimator (143) do not use. In this case, the quality of the shipping material is estimated using a robust quality estimator (144).
These robust quality estimators shown in FIG. 2 require input for estimation. This required input and related measurements are not shown in FIG. The robust quality estimator is further calibrated online using valid real measurements. This required online or offline measurement is not shown in FIG.

  In the plan of FIG. 2, the controller (129) controls the composition of the blended raw material mixture via (130), controls the operating conditions of the hydrodesulfurization unit via (136), and further controls the amount of low sulfur feedstock. (137) to control the hydrodesulfurization unit operation so that a product (120) with a quality close to or equal to the desired quality (to avoid product over-quality) is obtained. Try to optimize. This controller decision is based on the estimated quality measured at (131), (142), (143) and (144). Additional inputs to this controller are the price of the ingredients (see equation above) and the desired product characteristics.

  A model reference quality estimator that is preferably used in the method of the present invention needs to be calibrated using a historical quality measure. The use of historical quality measures is not straightforward. In the present invention, for example, the time (inactivity time) between the time of knowing a certain performance of the step (c) product, for example the actual measurement of the sulfur level, and the time of this actual measurement is relatively long. When using such historical data to calibrate process models, other phenomena such as dynamics between quality estimator inputs (eg hydrodesulfurization conditions and hydrodesulfurization feed quality) and measured quality In addition, a phenomenon usually referred to as a process gain variation, that is, a deviation in the ratio between the input and the output must be considered.

  To deal with these undesired situations, a method in which a quality estimator can be used is a so-called steady state, i.e. when the composition, state and speed of the process fluid are uniform and constant at the inlet and outlet of the operation. It is customary to calibrate quality estimators. Such a calibration gives good results for the system to be monitored, but is still considered sub-optimal because it does not use useful dynamic (unsteady state) information. For this reason, calibration must wait until the process reaches a steady operating point. Furthermore, the presence of a steady state detector is required to know when calibration can begin.

  The following calibration method is used that can be used under non-stationary conditions. A model estimator that uses such a calibration method is referred to as a robust quality estimator in the present invention. The robust quality estimator of the present invention provides a more accurate and robust quality prediction and improves the performance of the quality management plan according to the present invention.

This improved automatic online calibration method
A) collecting raw process data,
B) processing the data collected in step A) with the process model to obtain a predicted value of gas oil quality;
C) processing the predicted value with a dynamic transfer function to produce two intermediate signals;
D) storing the two intermediate signals obtained in step C) in the history as a time function;
E) From the history, during the effective actual measurement of the gas oil quality, the minimum and maximum absolute values of these two intermediate signals within a time corresponding to the minimum and maximum specific dead time (these values are the minimum And defining the maximum prediction point),
F) calculating a deviation as the difference between the effective actual measured value and the region surrounded by the possible minimum and maximum predicted values obtained in step E);
G) Step I) is started if the absolute value of the deviation obtained in Step F) is 0, or Step H) is started if the absolute value of the deviation obtained in Step F) is greater than 0. ,
H) introducing the deviation into the process model, and I) repeating steps A) to H).
including.

  The process model calibrated in the present invention is preferably a so-called parametric model obtained from historical process data and gas oil quality measurements. Examples of such models include, for example, Introduction to linear regression analysis by Montgomery and Peck, Multiple Linear Regression as described in John Willy & Sons, 1992, Linear Systems & PhysiologyIncentiveSensitiveInstrumentalPrince , 1980 linear dynamic model (in Laplace transform domain), and T. Poggio and F.M. Radial bias function network as described in Girosi, Network for application and learning, The Proceedings of the IEEE, 78 (9): 1485-1497, September 1990. Has a function). Those skilled in the art will select the type of process model for gas oil quality that best fits the goals they understand, depending on the nature of the process model used and the type of raw material data received.

  FIG. 3 shows a process model (1) that receives input from raw process data (2). The process model (1) provides an estimated gas oil quality (11) that is used as an input for a controller (12) that can control, for example, a valve (not shown). FIG. 3 also shows a module (3) that performs steps (C) and (D). In addition, a validation module (5) is presented that validates the actual measurement (4) of gas oil quality to obtain an effective actual measurement (6) of gas oil quality. Based on the input from module (3) and the effective actual measurement (6) of gas oil quality, the deviation is calculated in (7). If this deviation exceeds 0 as described in step (G), deviation (8) is preferably used for calibration of process model (1), utilizing Kalman filter (9).

  Collection of raw process data (2) in step (A) used in the method of the present invention can be performed by methods known in the art. In process control engineering, it is customary to measure data (2) at a number of points for a fixed time. For example, in refinery operations, operating parameters such as temperature, pressure and flow are usually measured at frequent intervals or in a continuous manner and stored and processed in many ways known to those skilled in the art.

  In order to obtain the predicted value of the gas oil quality (11) from the collected raw process data (2), the process model (1) described above is used in step (B). Therefore, process (B) is a gas oil quality prediction process.

  Step (C) is a step necessary for the automatic online calibration method. This step and subsequent steps will also be described with reference to FIG. In these steps, calculations are made of the minimum and maximum predicted values possible during an effective actual measurement of gas oil quality. Step (C) is preferably by applying two dynamic transfer functions (so-called uncertainty mechanics) to the predicted value (non-delayed real time) of gas oil quality (11) to produce two intermediate signals. Do. Dynamic transfer functions are described in a manner well known to those skilled in the art, for example, in Linear Systems, Prentice-Hall, Information & Systems sciences series, 1980 by Keilath. In step (D), these intermediate signals (20, 21) are stored in the history as a function of time. This necessarily results in an (uncertain) region (22) where the actual process response should be located, which becomes very narrow when the steady state situation (23, 24) is reached. In an unsteady state situation, the uncertainty region (22) can be reduced to a line corresponding to the case where the independent dynamic transfer function is the same (this situation is not shown in FIG. 4). is there. The so-called minimum and maximum prediction points are the absolute minimum values (27) and absolute maximum values (27) and absolute maximum values (27) of these two intermediate signals (20, 21) within a time corresponding to the specific dead time of minimum (25) and maximum (26). It is obtained by calculating 28). This dead time is calculated from the virtual position of the gas oil quality estimator relative to the actual measurement position of the gas oil quality, the measurement time of the gas oil actual quality, and other process conditions such as flow rate and liquid hold-up. It is a function. A person skilled in the art can easily measure the dead time. From this input, the maximum non-operation time (26) and the minimum non-operation time (25) are obtained from the effective actual measurement value (29 → 29 ′) of the gas oil quality in the step (F) and the predicted gas oil quality region (22 ) As well as the time of the process history to compare with the gas oil quality values at specific minimum (27) and maximum (28) points.

  Before reaching the steady state situation, the region (22) can change significantly. The state of the conventional system may simply be calibrated during steady state, or if a valid actual measurement of gas oil quality is in the region, it may cause incorrect calibration. However, the method of the present invention is designed in detail to calibrate only when a valid actual measurement (29) of gas oil quality is outside the uncertainty region (22), thereby preventing instability in the closed loop. ing. The calibration method of the present invention can be advantageously performed under steady state and unsteady state conditions.

  In step (E) of the method according to the invention, part of the calibration process consists of the deviation (30) (so-called prediction error), the valid actual measurement (29 ′) and the minimum point obtained in the previous calculation (27 ) And the distance between the regions (22) enclosed by the maximum point (28).

  The later validated actual measurement (29) of gas oil quality may be an online or offline measurement of gas oil quality. Examples of possible measurement methods including gas oil quality and near infrared and / or NMR are as described above.

  In step (G), the usefulness of the gas oil quality for effective calibration purposes is measured. Only gas oil quality measurements (29 ') outside the uncertainty region (22) can be used to calibrate the model. In other words, the absolute value of the deviation is zero in the calculation of the deviation (30) as described above, that is, the effective actual measurement of the gas oil is within the uncertainty region (22), more precisely, the minimum value of the gas oil quality value. If zero, which means between point (27) and maximum point (28), the measured deviation (30) is not used as further input for the calibration process and the system does not need to be improved. Therefore, the process performed up to now is repeated. However, if the calculated deviation (30) exceeds 0 in absolute value of the deviation (30) as shown in FIG. 4, the obtained deviation (30) is incorporated into the process model in step (H). Repeat the previous steps (step I). The final result is the emergence of a more accurate modification prediction process model and serves as a basis for further modification depending on the deviation level observed during the calibration process.

  Step (H) is preferably performed so that the Kalman filter (9) can be used to incorporate the deviation (8) into the process model (1) (see FIG. 3). As a result of performing the step (H) by such a method, the deviation parameter can be incorporated into the process model by adjusting the linear parameter of the deviation to update the prediction band and improving the process model. The use of Kalman filters is well known in the art of process control operations. In this regard, see “Stochastic Progressions and Filtering Theory” (Academic Press, Mathematical Sciences and Science and Engineering, Vol. 64, 1970) by Jazwinski. Since the Kalman filter is an essentially optimal stochastic filter, it is very suitable for use in the method of the present invention, filtering and even removing noise related to measured gas oil quality.

It should be noted that the use of the Kalman filter is not limited to calibration operations performed under non-steady state conditions, as can provide useful information when the process is performed under steady state conditions.
It has been found that an even more robust control method can be obtained by combining the method of the present invention with a Kalman filter. A further advantage of using a Kalman filter is that it maintains an improved accuracy of the gas oil quality estimation method. If a valid actual measurement of gas oil quality is not received, calibration as defined in steps E, F, G is not performed. The system repeats steps AD until a further effective actual measurement of gas oil quality is received.

The state of hydrodesulfurization operation technology is shown. 2 illustrates an embodiment having a preferred control plan of the present invention. A robust quality estimator is shown. A robust quality estimator is shown.

Explanation of symbols

101 Hydrodesulfurization unit 102, 103, 104 Raw material 105, 106, 107 Storage container 111 Compound storage 108, 109, 110 Container 112, 113 Storage container 114 Online additive injection 120 Final gas oil product 121 Kerosene feed 122 Direct Distillate oil feedstock 123 Vacuum gas oil feedstock 124 Circulating oil feedstock 125 Compound feedstock mixture 126, 127 Storage container 129 Expanded model process controller 130 Control line 131 Robust quality estimator 132 Hydrodesulfurization unit 133 Hydrodesulfurization unit Effluent 134 Low sulfur gas oil feed 135 Feedstock storage container 139 Additive storage container 140 Final gas oil product storage container or storage tank 141 Ship 142, 143, 144 Robust quality estimator 1 Process model 2 Raw process 3 Module 4 Actual measured value of gas oil quality 5 Activation module 6 Effective actual measured value of gas oil quality 7 Deviation calculation 8 Deviation 9 Kalman filter 11 Estimated gas oil quality 12 Controller 20, 21 Intermediate signal 22 Uncertain area or Predicted gas oil quality area 23, 24 Steady state situation 25 Minimum prediction point or minimum dead time 26 Maximum predicted point or maximum dead time 27 Absolute minimum 28 Absolute maximum 29 Activated after gas oil quality Actual measurement 29 'Effective actual measurement of gas oil quality 30 Deviation

Claims (20)

Starting from two or more hydrocarbon feedstocks with a sulfur content above 0.05% by weight, continuously producing hydrocarbon products with a specific value with a desired sulfur content of less than 0.05% by weight In the way to
(A) a step of blending two or more hydrocarbon feeds having a sulfur content exceeding 0.05% by weight to form a blended raw material mixture;
(B) a step of reducing the sulfur content of the blended raw material mixture in the hydrodesulfurization step;
(C) After obtaining the hydrocarbon fraction with reduced sulfur content, including the effluent of step (b), measuring the sulfur content of the hydrocarbon fraction, and (d) direct production of step (c) Obtaining a final hydrocarbon product from the product, said step comparing the sulfur content measured in step (c) with the desired sulfur content of the hydrocarbon product, Adjusting the process until the content approximates or is equal to the desired sulfur content of the hydrocarbon product,
Controlling the production of a hydrocarbon product having the desired sulfur content by integrated control of blending in step (a) and hydrodesulfurization unit operation in step (b), said integrated control in step (c) The addition of the low sulfur hydrocarbon feed to the effluent of step (b) is also controlled and the sulfur content of the low sulfur hydrocarbon feed mixed with the effluent of step (b) in step (c) is zero. The process for continuously producing a hydrocarbon product according to claim 5, wherein the sulfur content of the gas oil fraction in step (c) is measured after the addition of the low sulfur hydrocarbon feedstock.   The method according to claim 1, wherein a multivariable predictive controller is used for the integrated control in step (d).   In step (c), the sulfur content in the fraction is measured with a model-based quality estimator, and the sulfur content of the blended raw material mixture obtained in step (a) is determined to be sulfur in the fraction in step (c). The method according to claim 1, wherein the method is applied as information to the model-based quality estimator for predicting the content.   At least one performance other than the sulfur content of the fluid selected from the group consisting of the blended raw material mixture, the effluent of step (b), and the hydrocarbon fraction is also provided as other information to the model-based quality estimator. The method of claim 3.   4. The method of any one of claims 1-3, wherein the control of the process is arranged to bring at least one performance other than the sulfur content of the hydrocarbon product within the specifications of the hydrocarbon product. Method.   Said at least one other performance is cetane index, cetane number, cloud point, normal oil oil clogging point, flash point, pour point, density, viscosity, color tone, lubricity, electrical conductivity, total aromatic content, diene Claims selected from the group consisting of + -aromatic content, poly-aromatic content, distillation temperature at 90%, 95% or 100% recovery, distillation curve, distribution by sulfur species boiling range, and nitrogen content Item 6. The method according to Item 5.   The method according to claim 1, wherein the integrated control also controls the addition of additives in step (c).   The low sulfur hydrocarbon feedstock is selected from the group consisting of a gas oil fraction obtained from a fuel hydrocracker, a hydrotreated kerosene fraction, a fatty acid methyl ether, and a fuel fraction obtained from a Fischer-Tropsch reaction product The method of claim 1, wherein: Sulfur content of the blended feed mixture, the sulfur content of step (b) the effluent, the measurement of one performance selected from sulfur content by Li Cheng group of a hydrocarbon fraction, claim in which one measuring method is used The method according to any one of 1 to 8. Sulfur content of the blended feed mixture, the sulfur content of step (b) the effluent, the measurement of one performance selected from sulfur content by Li Cheng group of a hydrocarbon fraction, the quality estimator model criteria are used The method according to any one of claims 1 to 9. 11. A method according to claim 3, 4 or 10, wherein the model reference quality estimator is a self-calibrating quality estimator that can be calibrated under steady or unsteady state conditions. The model reference quality estimator obtains a sulfur content measurement using off-line or on-line X-ray fluorescence analysis or off-line or on-line ultraviolet fluorescence analysis and uses the effective actual measurement of this sulfur content to 11. A method according to claim 3, 4 or 10 calibrated in   13. A process according to any one of claims 1 to 12, wherein the hydrocarbon product is obtained in step (d) in a storage vessel filled with the hydrocarbon fraction of step (c).   The sulfur content in the storage vessel and / or the quality of at least one other performance is measured when partially filling the storage vessel, and the multivariable predictive controller also determines when the storage vessel is filled to a predetermined level. 14. The method of claim 13, wherein this measurement is utilized to adjust the method to control the performance of the total hydrocarbon product in the vessel.   15. The method of claim 14, wherein the measurement of sulfur content and / or at least one other performance quality is performed utilizing a model-based quality estimator.   The method according to any one of claims 1 to 15, wherein the hydrocarbon product is a gas oil product.   The hydrocarbon feedstock used in step (a) is a kerosene fraction, straight-run gas oil fraction, vacuum gas oil fraction, gas oil fraction obtained by pyrolysis, and light or heavy circulation obtained by fluid catalytic cracking 17. A method according to any one of the preceding claims comprising two or more refinery sources selected from a list of oils.   18. A process according to any one of claims 1 to 17, wherein step (b) is carried out in the presence of hydrogen and a heterogeneous catalyst containing a support, a Group VIB metal and a Group VIII non-noble metal.   The method according to any one of claims 1 to 18, wherein a catalytic dewaxing step is also performed in the step (b) after the hydrodesulfurization step. The calibration is
A) collecting raw process data,
B) processing the data collected in step A) with the process model to obtain a predicted value of gas oil quality;
C) processing the predicted value with a dynamic transfer function to produce two intermediate signals;
D) storing the two intermediate signals obtained in step C) as a history as a time function;
E) From the history, during the effective actual measurement of the gas oil quality, the minimum and maximum absolute values of these two intermediate signals within a time corresponding to the minimum and maximum specific dead time (these values are possible Defining minimum and maximum prediction points)
F) calculating a deviation as the difference between the effective actual measured value and the region surrounded by the possible minimum and maximum predicted values obtained in step E);
G) If the absolute value of the deviation obtained in step F) is 0, proceed to step I), or if the absolute value of the deviation obtained in step F) is greater than 0, proceed to step H).
H) introducing the deviation into the process model, and I) repeating steps A) to H).
The method of claim 11 performed by:

Images